Eclipse Community Forums
Forum Search:

Search      Help    Register    Login    Home
Home » Eclipse Projects » Java WorkFlow Tooling (JWT) » Has anyone looked into Imixs. <Open Source Workflow/> .org
Has anyone looked into Imixs. <Open Source Workflow/> .org [message #10241] Mon, 09 January 2006 21:22 Go to next message
Charles Fulnecky is currently offline Charles Fulnecky
Messages: 81
Registered: July 2009
Member
http://www.imixs.org/websites/imixs-org_en.nsf/chapter/0030. ?OpenDocument

Just curious what others in this group have to say about it.

Obviously it's not EMF/GMF, so what are the relative advantages to using
EMF and or GMF if any?

Thanks,

Charlie
Re: Has anyone looked into Imixs. <Open Source Workflow/> .org [message #10308 is a reply to message #10241] Tue, 10 January 2006 04:19 Go to previous messageGo to next message
Eclipse User
Originally posted by: fabrice.dewasmes.openwide.fr

Charlie wrote:
> http://www.imixs.org/websites/imixs-org_en.nsf/chapter/0030. ?OpenDocument
>
> Just curious what others in this group have to say about it.
>
> Obviously it's not EMF/GMF, so what are the relative advantages to using
> EMF and or GMF if any?

I didn't know about this project and it seems interesting. However they
use a 'proprietary' (meaning non standard) workflow model. But this is
interesting and the UI seems pretty clean ! Looking at javadoc shows
that they do not use EMF but only GEF. This is the approach we have used
in our early implementation.

To me, the interest of using EMF is :
- generating classes that have nearly no value add
- persistence of the model instances. This is the most interesting part
especially when you use an XSD schema and you want to generate XML
instances that conforms to this schema. This avoids use of direct XML
manipulation or Object-XML mapping such as XMLBeans or JAXB (even if I
find XMLBeans pretty useful).

This alone is not sufficient to switch to EMF. I think the real interest
of using GEF would be to also use GMF. But at this time the project
seems not sufficiently stable to use it... I will give it a new try
using your pointers about versions to use.

Fabrice

>
> Thanks,
>
> Charlie
>
>
rewrite the model? [message #10341 is a reply to message #10308] Tue, 10 January 2006 23:44 Go to previous messageGo to next message
Eclipse User
Originally posted by: xxp1978.163.com

using emf or gmf ?
plan to rewrite the XPDL Model?
u asked, I answer by IMHO [message #10672 is a reply to message #10308] Thu, 12 January 2006 07:25 Go to previous messageGo to next message
Eclipse User
Originally posted by: vgv8.yahoo.com.br

I am just fan of .NET and in comparison I see clearly many things are bad to
me as user in GEF with EMF (and GMF as the bridge bewtween them). And from
developing the wrong things you get huge wrong things

I just cannot restrict me from some comments:
- inixs.com is IBM Business partner and do not use EMF;
- GMF is possibly tapping the problems on top of somewhat excessively
ambiguous and exaggerated initial ideas, arquitectures and thereafter
problems (GEF with EMF),
it may possibly just insist on wrong arquitectures and vision
- I hate the idea that Swing as proven, elaborated MVC solution with a lot
of ready solutions is somewhat incompatible with Eclipse platform.
- - As a matter of fact, I think the attitude to melt the platform and
IDE was bad concept )oSGI tries to revert it partially but it resemles me
shooting out your own legs)

Might be it had more sense just to have more simple or integrated with
IDE/platform solutions. But the problems are also in platform and IDE.

Guennadi Vanine
"Fabrice Dewasmes" <fabrice.dewasmes@openwide.fr> wrote in message
news:dpvua1$lsc$1@utils.eclipse.org...
> Charlie wrote:
>> http://www.imixs.org/websites/imixs-org_en.nsf/chapter/0030. ?OpenDocument
>>
>> Just curious what others in this group have to say about it.
>>
>> Obviously it's not EMF/GMF, so what are the relative advantages to using
>> EMF and or GMF if any?
>
> I didn't know about this project and it seems interesting. However they
> use a 'proprietary' (meaning non standard) workflow model. But this is
> interesting and the UI seems pretty clean ! Looking at javadoc shows that
> they do not use EMF but only GEF. This is the approach we have used in our
> early implementation.
>
> To me, the interest of using EMF is :
> - generating classes that have nearly no value add
> - persistence of the model instances. This is the most interesting part
> especially when you use an XSD schema and you want to generate XML
> instances that conforms to this schema. This avoids use of direct XML
> manipulation or Object-XML mapping such as XMLBeans or JAXB (even if I
> find XMLBeans pretty useful).
>
> This alone is not sufficient to switch to EMF. I think the real interest
> of using GEF would be to also use GMF. But at this time the project seems
> not sufficiently stable to use it... I will give it a new try using your
> pointers about versions to use.
>
> Fabrice
>
>>
>> Thanks,
>>
>> Charlie
Re: u asked, I answer by IMHO [message #13185 is a reply to message #10672] Fri, 13 January 2006 05:46 Go to previous messageGo to next message
Eclipse User
Originally posted by: fabrice.dewasmes.openwide.fr

Guennadi Vanine wrote:
> I am just fan of .NET and in comparison I see clearly many things are bad to
> me as user in GEF with EMF (and GMF as the bridge bewtween them). And from
> developing the wrong things you get huge wrong things
>
> I just cannot restrict me from some comments:
> - inixs.com is IBM Business partner and do not use EMF;
I noticed this.
> - GMF is possibly tapping the problems on top of somewhat excessively
> ambiguous and exaggerated initial ideas, arquitectures and thereafter
> problems (GEF with EMF),
> it may possibly just insist on wrong arquitectures and vision
> - I hate the idea that Swing as proven, elaborated MVC solution with a lot
> of ready solutions is somewhat incompatible with Eclipse platform.
> - - As a matter of fact, I think the attitude to melt the platform and
> IDE was bad concept )oSGI tries to revert it partially but it resemles me
> shooting out your own legs)

Participating in an eclipse plugin is about writing a set of frameworks
or tools, plugins etc... for eclipse not rewriting eclipse ;) As I've
said using eclipse swing bridge is not a good idea.

Fabrice

>
> Might be it had more sense just to have more simple or integrated with
> IDE/platform solutions. But the problems are also in platform and IDE.
>
> Guennadi Vanine
> "Fabrice Dewasmes" <fabrice.dewasmes@openwide.fr> wrote in message
> news:dpvua1$lsc$1@utils.eclipse.org...
>
>>Charlie wrote:
>>
>>> http://www.imixs.org/websites/imixs-org_en.nsf/chapter/0030. ?OpenDocument
>>>
>>>Just curious what others in this group have to say about it.
>>>
>>>Obviously it's not EMF/GMF, so what are the relative advantages to using
>>>EMF and or GMF if any?
>>
>>I didn't know about this project and it seems interesting. However they
>>use a 'proprietary' (meaning non standard) workflow model. But this is
>>interesting and the UI seems pretty clean ! Looking at javadoc shows that
>>they do not use EMF but only GEF. This is the approach we have used in our
>>early implementation.
>>
>>To me, the interest of using EMF is :
>>- generating classes that have nearly no value add
>>- persistence of the model instances. This is the most interesting part
>>especially when you use an XSD schema and you want to generate XML
>>instances that conforms to this schema. This avoids use of direct XML
>>manipulation or Object-XML mapping such as XMLBeans or JAXB (even if I
>>find XMLBeans pretty useful).
>>
>>This alone is not sufficient to switch to EMF. I think the real interest
>>of using GEF would be to also use GMF. But at this time the project seems
>>not sufficiently stable to use it... I will give it a new try using your
>>pointers about versions to use.
>>
>>Fabrice
>>
>>
>>>Thanks,
>>>
>>>Charlie
>
>
>
on bridges and gaps [message #13324 is a reply to message #13185] Fri, 13 January 2006 09:27 Go to previous messageGo to next message
Eclipse User
Originally posted by: vgv8.yahoo.com.br

> Participating in an eclipse plugin is about writing a set of frameworks or
> tools, plugins etc... for eclipse not rewriting eclipse ;) As I've said
> using eclipse swing bridge is not a good idea.
>
> Fabrice

IMHO using Eclipse Swing Bridge is no worse idea than using Eclipse or
making bridge between Sun and IBM.
It is just compromise and dialectics:
should this world have bridges or only straight gaps?

Guennadi Vanine
I need clarifications [message #13425 is a reply to message #10308] Fri, 13 January 2006 17:40 Go to previous messageGo to next message
Eclipse User
Originally posted by: vgv8.yahoo.com.br

Fabrice:
I should tell that I had not been looking into GEF with EMF for more than a
month
(since I shifted back to Vision,Requirements basics and was out for another
project).
I should repeat that my interest is adjust XML format to a proprietory
format and fix it. Since BPM engine is considered fixed or , at least, not
available to my consideration.
That is only Workflow designer with fixed rigid necessities.

First of all, there are 2 kind of friendlinesses:
- user-friendliness for user or biz analyst (that doesn't know anything
about programming - formats, standards, XML, underlying frameworks, Java)
- developer friendliness

As for developer friendliness I may tell, that it is impossible in my
context to specify all workflow only from client side:
- hierarchy of users are specified in the database (by data modeling) or in
LDAP;
- from the other side, current workflow designer (and therefore a new one I
recreate) permits Java scripting
http://www.beanshell.org/
It is again is not for END-users.

Developer-friendliness targets possibility to have flexibility to change the
compiled designer easily

From the other side there is no need for my designer in changing workflow
models and XML formats.

If to have more ambitious opensource multipurpose project interoperable
project like JWT you MISSED to note the main developer-friendliness or even
user-friendliness:
- it is ability to specify workflow models and this is somewhat
orthogonal/independent on graphic representations and XML formats

I couldn't understand your phrases:
> To me, the interest of using EMF is :
> - generating classes that have nearly no value add
What is the interest in thing without "value add"?
> - persistence of the model instances. This is the most interesting part
What do you mean under model instances?
For me the problem is that:
- compiled product (user-friendliness) should read definitions of Workflow
and for Workflow Nodes for sidebar to insert in workflow inside the same
defintion of Workflow model (in GEF with EMF there are really EMF ECore
metamodel to 4 others models in XMI, Java, etc. that should be syncronized);
- for developer ("developer-friendliness") I should easily redefine models
or read it from a file and recompile product

> especially when you use an XSD schema and you want to generate XML
This was what I missed in GEF with EMF. If I specify models through EMF
ECore metamodel (and then recieve models in Java, XMI, etc.), I should do
all through EMF ECore model.
And my requirement is to introduce/adjust my own XML/XMI format according to
my XSD.
How do I change it?

I didn't have much time to understand it but as I understood EMF XMI is
given? How can I change XSD for EMF XMI and for EMF ECore model?

It is also the same problem as why, for ex., MyEclipse (plugin) doesn't
recognize EMF XMI.
http://www.myeclipseide.com/PNphpBB2+file-viewtopic-t-10320. html

Let me remind that MyEclipse permits UML, dataabase modeling,
reverse-engineering (but not with EMF formats)

Interoperability with tools! Isn't it JWT objective. As a matter of fact I
even do not know which free tools work with EMF!!! See my earlier posts on
having problems with OmondoEclipseUML
http://www.forum-omondo.com/viewtopic.php?t=827

> instances that conforms to this schema. This avoids use of direct XML
Yeeeah. I have contrary reuirement. Nobody wants to study EMF and would like
to redefine everything in kust XML
> manipulation or Object-XML mapping such as XMLBeans or JAXB (even if I
> find XMLBeans pretty useful).

Can you give any links on Object-XML mapping and ECore XMI redefinition
according to XSD
It all seemed to me as the problems for other frameworks and projects
outside of GEF and EMF

>
> This alone is not sufficient to switch to EMF. I think the real interest
There a lot of very interesting things in the world (and I have only 24 H in
my day)
> of using GEF would be to also use GMF. But at this time the project
GMF to me is, so far, is not even conceptual issues. It is a tool for GEF
with EMF work
(that IMHO doesn't address many conceptual issues)
> seems not sufficiently stable to use it... I will give it a new try using
> your pointers about versions to use.
me too

Guennadi Vanine
impossible to specify (dynamic) behaviours thru static modeling tools [message #15118 is a reply to message #13425] Fri, 13 January 2006 18:03 Go to previous messageGo to next message
Eclipse User
Originally posted by: vgv8.yahoo.com.br

The problem with EMF XSD is that it specifies static models throgh
class-types diagrams.

But how shall I attach to GEF with EMF some dynamic elements?
How shall I change at tun-time XSD? I do not know what user will enter and
cannot specify it thru model.

I need to attach to graphical elements (workflow nodes) specified thru model
some datastructures of arbitrary format at runtime.

I have graphical elements - workflow nodes.
And I need to double click it in a datagrid windows (or in property view)
add data and it should be incorporated inside XML of workflow elements
definitions but at run-time.
How shall I incorporate this into EMF XML?

I have a reuirement to have final XML representation to be fused in one XML
before deployment to server side BPM engine.

One of the thought (amongst other) that I have is to produce the EMF ECore
model once and then just use the produced Java code for manual hacking

This problem is quite similar to the problem of incompatibility with Swing
use.

Any modeling tools, frameworks are for static specification, some predefined
during modeling carcass but I do not know any application without dynamic
behaviour.

Guennadi Vanine

"Guennadi Vanine" <vgv8@yahoo.com.br> wrote in message
news:dq9a8n$pa2$1@utils.eclipse.org...
> Fabrice:
> I should tell that I had not been looking into GEF with EMF for more than
> a month
> (since I shifted back to Vision,Requirements basics and was out for
> another project).
> I should repeat that my interest is adjust XML format to a proprietory
> format and fix it. Since BPM engine is considered fixed or , at least, not
> available to my consideration.
> That is only Workflow designer with fixed rigid necessities.
>
> First of all, there are 2 kind of friendlinesses:
> - user-friendliness for user or biz analyst (that doesn't know anything
> about programming - formats, standards, XML, underlying frameworks, Java)
> - developer friendliness
>
> As for developer friendliness I may tell, that it is impossible in my
> context to specify all workflow only from client side:
> - hierarchy of users are specified in the database (by data modeling) or
> in LDAP;
> - from the other side, current workflow designer (and therefore a new one
> I recreate) permits Java scripting
> http://www.beanshell.org/
> It is again is not for END-users.
>
> Developer-friendliness targets possibility to have flexibility to change
> the compiled designer easily
>
> From the other side there is no need for my designer in changing workflow
> models and XML formats.
>
> If to have more ambitious opensource multipurpose project interoperable
> project like JWT you MISSED to note the main developer-friendliness or
> even user-friendliness:
> - it is ability to specify workflow models and this is somewhat
> orthogonal/independent on graphic representations and XML formats
>
> I couldn't understand your phrases:
>> To me, the interest of using EMF is :
>> - generating classes that have nearly no value add
> What is the interest in thing without "value add"?
>> - persistence of the model instances. This is the most interesting part
> What do you mean under model instances?
> For me the problem is that:
> - compiled product (user-friendliness) should read definitions of Workflow
> and for Workflow Nodes for sidebar to insert in workflow inside the same
> defintion of Workflow model (in GEF with EMF there are really EMF ECore
> metamodel to 4 others models in XMI, Java, etc. that should be
> syncronized);
> - for developer ("developer-friendliness") I should easily redefine models
> or read it from a file and recompile product
>
>> especially when you use an XSD schema and you want to generate XML
> This was what I missed in GEF with EMF. If I specify models through EMF
> ECore metamodel (and then recieve models in Java, XMI, etc.), I should do
> all through EMF ECore model.
> And my requirement is to introduce/adjust my own XML/XMI format according
> to my XSD.
> How do I change it?
>
> I didn't have much time to understand it but as I understood EMF XMI is
> given? How can I change XSD for EMF XMI and for EMF ECore model?
>
> It is also the same problem as why, for ex., MyEclipse (plugin) doesn't
> recognize EMF XMI.
> http://www.myeclipseide.com/PNphpBB2+file-viewtopic-t-10320. html
>
> Let me remind that MyEclipse permits UML, dataabase modeling,
> reverse-engineering (but not with EMF formats)
>
> Interoperability with tools! Isn't it JWT objective. As a matter of fact I
> even do not know which free tools work with EMF!!! See my earlier posts on
> having problems with OmondoEclipseUML
> http://www.forum-omondo.com/viewtopic.php?t=827
>
>> instances that conforms to this schema. This avoids use of direct XML
> Yeeeah. I have contrary reuirement. Nobody wants to study EMF and would
> like to redefine everything in kust XML
>> manipulation or Object-XML mapping such as XMLBeans or JAXB (even if I
>> find XMLBeans pretty useful).
>
> Can you give any links on Object-XML mapping and ECore XMI redefinition
> according to XSD
> It all seemed to me as the problems for other frameworks and projects
> outside of GEF and EMF
>
>>
>> This alone is not sufficient to switch to EMF. I think the real interest
> There a lot of very interesting things in the world (and I have only 24 H
> in my day)
>> of using GEF would be to also use GMF. But at this time the project
> GMF to me is, so far, is not even conceptual issues. It is a tool for GEF
> with EMF work
> (that IMHO doesn't address many conceptual issues)
>> seems not sufficiently stable to use it... I will give it a new try using
>> your pointers about versions to use.
> me too
>
> Guennadi Vanine
>
static vs dynamic data [message #15139 is a reply to message #15118] Fri, 13 January 2006 18:37 Go to previous messageGo to next message
Eclipse User
Originally posted by: vgv8.yahoo.com.br

Once more.

Any Workflow Designer should permit adding data at run-time according to
somewhat in end-user mind (it is not just editing graphics).

EMF with GEF is very good to specify a priory known during modeling behavior
and structure of purely graphical elements specified at modeling time.
When you redefine the EMF ECore model u should resyncronize, regenerate from
it Java, XMI, XSD models.
But how do someone attach to statically specified structures and behaviours
dynamic data at run-time?
But it doesn't help and even to the contrary to run-time necessities

I should somehow:
- separate static, definable during modeling structures (workflow nodes
coordinates, shapes, etc.),
- attach to them at run-time dynamic data,
- fuse both in one XML file for deployment
- upload, deploy to server BPM engine
and permit the contrary order:
- download workflow definition
- distinguish pure graphic and semantic data for further editing

If GEF with EMF helps treating just visual (static or known during modeling)
data, with pure graphical editors, workflow editors should attach unknown
beforehand data

Might be I miss something but for workflow editors, EMF arquitecture
complicate instead of helping things.
At least, I could not find any ready answers to these problems. I just
should study other frameworks, projects. But I am not sure that EMF is right
architecture at all for this (though it depends, as I told many times, on
resources and scope of project. Mine is not to define frameworks but create
specific application)!

You may contact me for on-line discussion at:
MSN Messenger: VGV888<AT>HOTMAIL.com
YAHOO IM: VGV8

Guennadi Vanine
I looked in [message #15172 is a reply to message #10241] Fri, 13 January 2006 21:52 Go to previous messageGo to next message
Eclipse User
Originally posted by: vgv8.yahoo.com.br

Hi, Carhlie and all,
I tried to look in IX Workflow Modeler (upon your pointing)
but couldn't download it (the links are broken)

I couldn't get download, then the forum has 2 posts for 1.5 years
(1 visitor question and one reply)

My dream to be the first or second one came true, I am the second registered
user there.
Look for more details in my post there in forum.

But it is not exactly what I need - I need associate to graphical elements
(Workflow Nodes) a data (by Datagrid, Property View) at RUNTIME and
incorporate it in the same XML-file describing workflow. That simple in
Swing and that complicated with EMF.

I am just impressed by the volume of the OpenSource projects, their support
and absence of any interest. Kind of decentralization, desintegration that
jump into eyes after .NET flourishing communities.

Guennadi Vanine

"Charlie" <cafulnecky@yahoo.com> wrote in message
news:dpv5pg$g5s$1@utils.eclipse.org...
> http://www.imixs.org/websites/imixs-org_en.nsf/chapter/0030. ?OpenDocument
>
> Just curious what others in this group have to say about it.
>
> Obviously it's not EMF/GMF, so what are the relative advantages to using
> EMF and or GMF if any?
>
> Thanks,
>
> Charlie
>
Re: impossible to specify (dynamic) behaviours thru static modeling tools [message #15377 is a reply to message #15118] Sat, 14 January 2006 21:35 Go to previous messageGo to next message
Charles Fulnecky is currently offline Charles Fulnecky
Messages: 81
Registered: July 2009
Member
Guennadi,

See http://www.devx.com/Java/Article/29093/0/page/2 and 3 for some
interesting concepts.
What you really need is a tool that is both a tool to create the "class"
definitions and create "class instances" populated with data. The real
trick is to determine a graphical representation that is generic enough to
display your "instances".



Charlie

"Guennadi Vanine" <vgv8@yahoo.com.br> wrote in message
news:dq9bkh$r7f$1@utils.eclipse.org...
> The problem with EMF XSD is that it specifies static models throgh
> class-types diagrams.

I think you are still confusing Omondo's UML tool with EMF. Omondo is just
one of may ways to build ecore.

>
> But how shall I attach to GEF with EMF some dynamic elements?
> How shall I change at tun-time XSD? I do not know what user will enter and
> cannot specify it thru model.

You let the user create the "class" graphically (as in Omondo) using
something simple like the class diagram ... you have to come up with some
sort of generic representation for the populated "instance" of course. The
above link should give you some ideas on how you can serialize the resulting
schema and it's corresponding xml, probably in a 2 step process.

>
> I need to attach to graphical elements (workflow nodes) specified thru
> model some datastructures of arbitrary format at runtime.
>
> I have graphical elements - workflow nodes.
> And I need to double click it in a datagrid windows (or in property view)
> add data and it should be incorporated inside XML of workflow elements
> definitions but at run-time.
> How shall I incorporate this into EMF XML?
>
> I have a reuirement to have final XML representation to be fused in one
> XML before deployment to server side BPM engine.
>
> One of the thought (amongst other) that I have is to produce the EMF ECore
> model once and then just use the produced Java code for manual hacking
>
> This problem is quite similar to the problem of incompatibility with Swing
> use.
>
> Any modeling tools, frameworks are for static specification, some
> predefined during modeling carcass but I do not know any application
> without dynamic behaviour.
>
> Guennadi Vanine
and where is the trick? [message #15476 is a reply to message #15377] Sun, 15 January 2006 10:19 Go to previous message
Eclipse User
Originally posted by: vgv8.yahoo.com.br

Hi, Charlie,
Here is conceptual problem with any modeling.

It is impossible to add to statical models (specified through
"class-diagram" type static) behavioral artifacts. No such tools were
invented conceptually.

And even for such static modeling in EMF, there are no free integrated
trustworthy tools.

I do not see how to make a trick with EMF. I cannot restrict a user to some
GENERIC template for adding data at run-time and then fuse it in one XML
with pure graphical data (on shapes).

Guennadi Vanine
Previous Topic:proposal: compactable workflow parts
Next Topic:Relationships with other Eclipse projects
Goto Forum:
  


Current Time: Mon Jul 28 12:29:40 EDT 2014

Powered by FUDForum. Page generated in 0.02981 seconds